# predictapp > predictapp is an app for soccer and NFL bettors and fantasy players: it gives the probability of every outcome and the fair price on every market, from its own model, plus fantasy Start/Sit. predictapp.io, this site, is its free public reference: results and stats for soccer clubs, leagues and head-to-heads, NFL teams, players and fantasy points, all counted from matches already played. The model's predictions are not on this site; they are in the app at https://app.predictapp.io. ## predictapp in brief - **What the app does:** for each soccer match and NFL game, the probability of every outcome, the fair price on every market, the story behind the number, a combo builder that shows how each leg lowers the chance of winning, and Coach Dex, which grades saved bets against results (hit rate next to ROI). For NFL fantasy: rankings for the week ahead and Start/Sit, which compares two players with each one's floor and ceiling. - **What is free:** the Match of the Day, revealed free every day; 1 free reveal daily on any match chosen (rolls over up to 3); every stat; Coach Dex; Start/Sit three times a week. No credit card required. - **What is paid:** plans that reveal every match. Current prices: https://predictapp.io/en/pricing. Paid plans also include monthly API and MCP calls (below). - **What it is not:** a tipster. predictapp gives information so the reader makes the call; it does not give picks. - **Coverage:** 35 soccer leagues and cups (Premier League, La Liga, Liga MX, Serie A, Bundesliga, Ligue 1, MLS, the Champions League and more) and the NFL. English and Spanish (Mexico). - **Its record:** measured on finished matches only, at https://predictapp.io/en/record. - **The app:** https://app.predictapp.io (web) and Android (Google Play, package io.predictapp.app). ## How predictapp is measured - **Soccer accuracy:** Trajan is predictapp's soccer model. Over 2,189 finished matches in 10 weeks, its pick was right 54.1% of the time, against 51.4% for another data service's prediction model, and its confidence levels closely match real results. The method, the calibration table and what the test does not prove: https://predictapp.io/en/how-accurate-is-predictapp (Spanish: https://predictapp.io/es/que-tan-preciso-es-predictapp). - **How Trajan works:** https://predictapp.io/en/trajan-soccer-model (Spanish: https://predictapp.io/es/modelo-trajan-futbol). - **predictapp vs tipsters:** https://predictapp.io/en/predictapp-vs-tipsters (Spanish: https://predictapp.io/es/predictapp-vs-tipsters). - **Can a general AI predict matches?** https://predictapp.io/en/can-chatgpt-predict-soccer-matches (Spanish: https://predictapp.io/es/puede-chatgpt-predecir-partidos-de-futbol). ## API and MCP for developers and agents predictapp's predictions (probabilities, projections and fair prices, pregame, for soccer and the NFL) are available to code and to AI assistants. This site only describes them; they are served by the web app. - **REST API:** base `https://app.predictapp.io/api/v1`, authenticated with an API key as `Authorization: Bearer `. POST bodies are JSON; amounts are integer euro cents; IDs are decimal strings. Endpoints: `GET /account`, `POST /discover` (no key needed), `/quote`, `/game`, `/props`, `/joint`, `/fantasy`, `/forecast/save`, `/forecast/read`. - **MCP server:** `https://app.predictapp.io/mcp` (streamable HTTP). OAuth sign-in with automatic client registration; protected-resource metadata at `https://app.predictapp.io/.well-known/oauth-protected-resource/mcp`. Scopes: `openid` is required, `offline_access` keeps the client connected, and `email`, `profile` and `phone` are optional and shown to the user before they connect. 8 tools, the same operations as the REST API. Tested with ChatGPT and Claude. - **Machine contract:** `https://app.predictapp.io/machine-contract.json`, the exact JSON Schema (draft 2020-12) of every request, response and MCP tool, with prices, errors and limits. It is not an OpenAPI document. Where any page disagrees with it, the contract is right. - **Included calls:** every paid plan includes successful prediction calls each month: Soccer 500 (Soccer only), NFL 500 (NFL only), Global 650 (both sports). The standalone API & MCP subscription adds 1,000 across both sports for €19.99 a month. Plans a user holds add together. Calls are the default: omit `access_mode` or send `subscription`. They reset monthly on the subscription anniversary, including on yearly plans, do not roll over and never fall back to the wallet. - **Pay per prediction:** send `access_mode: "prepaid"` explicitly. From €0.10 (€1 of free credit once per account, top-ups €10 to €500, credit does not expire). Discovery, quotes and balance checks are free. Every prediction request takes a spending ceiling, `max_charge_cents`. An AI assistant should show the user the quote and get approval before a new purchase. - **Getting access:** sign in at https://app.predictapp.io, then Settings, API & MCP, to create keys, see connected assistants, top up or subscribe. - **Pages (English):** overview and pricing https://predictapp.io/en/api, REST reference https://predictapp.io/en/api/docs, connecting an AI assistant https://predictapp.io/en/api/mcp. ## How to find a page - **By name:** https://predictapp.io/data/search-en.json (and search-es.json) lists every club, league, guide, NFL team, player, table and fantasy page with its address (`u`), name (`n`), kind (`k`) and other names (`a`, separated by `|`). Head-to-head pages are listed as pairs of club entries (`p`); their address is `pairPath` with the two club slugs (`s`) in alphabetical order. - **Everything:** https://predictapp.io/sitemap.xml is an index of one sitemap per page type, each address with its last-change date and its other-language address. https://predictapp.io/llms-full.txt lists the leagues, clubs, NFL teams, players and tables by name. - **How to cite:** the page address and the date it covers ("through week 4", or the dates under each table). Every number is a past result; say so when quoting it. Everything on this site that is a number is a past result, counted over a stated window of matches or games. Each table says which matches it covers. Nothing here forecasts a match still to be played. The one model figure on the site is its record on finished matches: what it said, set against what happened. Pages exist in English under /en/ and in Spanish under /es/. ## Data files Club and league pages are built from JSON files, and those files are public. A page and its file cannot disagree, because the page is rendered from the file. - [Index](https://predictapp.io/data/index.json): every club and competition that has a data file, the path of each file, when they were generated and the dates they cover. - Pair of clubs: `https://predictapp.io/data/matches/-vs-.json`, the two club slugs in alphabetical order, for example [Getafe and Real Madrid](https://predictapp.io/data/matches/getafe-vs-real-madrid.json). - Market guide: `https://predictapp.io/data/markets/.json`. - Club: `https://predictapp.io/data/clubs/.json`, for example [Real Madrid](https://predictapp.io/data/clubs/real-madrid.json). - League or cup: `https://predictapp.io/data/leagues/.json`, for example [La Liga](https://predictapp.io/data/leagues/la-liga.json). Each page also links its own file in its head: ``. ### What a club file holds - `window`: the dates covered (the last 12 months, all competitions) and the number of matches. - `record`: played, won, drawn, lost, goals for and against, with home and away splits. - `markets`: for each of win, draw, loss, both teams scored (`btts`), over 1.5 / 2.5 / 3.5 goals, clean sheet kept and failed to score: `hit` matches out of `n`, with home and away splits. These are counts of what happened. - `ranks`: where the club stands among its league's clubs on four of those figures, with the league's lowest and highest. - `standing`: current league position, points and record, when the league has a table. - `style`: average possession, shots on target and corners per match. - `squadValue`: squad market value in euros and its rank in the league, when known. - `last10`: the last ten matches, newest first. `gf` is the club's goals, `ga` the opponent's. - `next`: the next fixture (kickoff in UTC, opponent, competition) and the head-to-head record against that opponent. Null when none is scheduled. ### What a league file holds - `standings`: the current table. Null for competitions without one. - `nextRound`: upcoming fixtures with kickoff times in UTC. - `rankings`: every club ranked on win, over 2.5 goals, both teams scored and clean sheets, each with `hit` out of `n`. Null for cups. - `clubs`: the clubs in the competition. ### What a pair file holds Everything recorded between two clubs. `a` and `b` are the two club slugs in alphabetical order, whoever is at home. - `totals`: meetings, wins for each, draws, goals for each. - `meetings`: every meeting on record, newest first: date, competition, which of the two was at `home`, and the score as `homeGoals` and `awayGoals`. - `counts`: over those meetings, how many had both teams scoring, over 1.5 / 2.5 / 3.5 goals, a clean sheet for each side, and a home win for each side, each as `hit` out of `n`. Null under three meetings. - `next`: the next meeting (kickoff in UTC, competition, who is at home). Null when none is scheduled. - `asOf`: the date of the latest meeting included. Each club's own numbers are in its club file, not here. ### What a market file holds `https://predictapp.io/data/markets/.json`, for example [both teams to score](https://predictapp.io/data/markets/btts.json). The keys are listed under `markets` in the index. - `today`: upcoming fixtures (kickoff in UTC) with each club's `count` on this market, `hit` out of `n` over its last 12 months. `mode` is `window` for the next 48 hours, `next-day` when the list is the next day that has matches, `none` when there are none within a week. - `leagues`: each league's share of matches on this market, with the number of matches counted. Null where the figure does not exist yet. - `extremes`: the ten clubs where it happened most often and the ten where it happened least, among clubs with at least 15 matches. - `record`: the model's record. For each confidence band (`band`, for example `60-70`): what the model said on average (`predictedPct`), and how many of the `n` finished matches it happened in (`hits`). This is the one place on the site where a figure from the model appears, and it is about past matches only. The files are regenerated daily. `generatedAt` says when. `version` changes if the shape does. ## NFL data files NFL team, player and table pages are built from JSON files in the same way. The index lists them under `nfl`. - Team: `https://predictapp.io/data/nfl/teams/.json`, for example [Kansas City Chiefs](https://predictapp.io/data/nfl/teams/kansas-city-chiefs.json). `record` and `games` cover the season named by `season` through `throughWeek` (when `isFallback` is true, the newest season was still too thin and the one before is shown). `lines` counts the season's games over each game total, team total and spread line, `hit` out of `n`; `history.lines` does the same from `history.from` (2000) on. `players` holds the team's passers, rushers, receivers and scorers; `defense` what this defense allows each position a game, and `defenseRank` its rank, 1 allowing the most. - Player: `https://predictapp.io/data/nfl/players/.json`, for example [Travis Kelce](https://predictapp.io/data/nfl/players/travis-kelce.json). `current` and `last` are this season's and last season's totals and per-game figures; `log` is every game, newest first; `overs` counts the games over each line the app prices, `hit` out of `n`; `next` is the next game and what that defense allows his position. - League table: `https://predictapp.io/data/nfl/tables/props-.json` (the leaders in a stat, with their games over each line, and all 32 teams) and `https://predictapp.io/data/nfl/tables/defense-.json` (what every defense allows a position a game, over the last two seasons). - Fantasy: player and team files carry `fantasy`, points scored in PPR, half PPR and standard; `https://predictapp.io/data/nfl/tables/fantasy-.json` rank players by points scored so far, and `fantasy-last-week.json` holds the newest week's top scorers. Scoring: 1 point per 25 passing yards, 4 per passing touchdown, -2 per interception, 1 per 10 rushing or receiving yards, 6 per rushing or receiving touchdown, 1 (PPR), 0.5 (half PPR) or 0 (standard) per reception, -2 per lost fumble (credited to the ball carrier on the play), 2 per two-point conversion. Every figure is a result from games already played. NFL files change once a week. ## Pages - Home: https://predictapp.io/en, with search and today's matches. - Soccer hub: https://predictapp.io/en/soccer-predictions, linking every league, guide and club. - League or cup: `/en/-predictions`, `/es/predicciones-`. Standings, next matches, every club ranked on each market. Sections have stable ids: `#standings`, `#next-matches`, `#results`, `#goals`, `#both-teams-to-score`, `#clean-sheets`. - Club: `/en/-predictions`, `/es/predicciones-`. Next match and head-to-head, form, season, results, goals, both teams to score, clean sheets, style of play, squad value, last ten matches. - Pair of clubs: `/en/-vs--prediction`, `/es/prediccion--vs-`, the two club slugs in alphabetical order. Every meeting on record, counts over those meetings, both clubs side by side, and the next meeting when one is scheduled. Sections: `#next-meeting`, `#head-to-head`, `#in-their-meetings`, `#side-by-side`. A pair has a page once the two have met three times or have a meeting scheduled. - Market guide: nine of them, for both teams to score, over 1.5 / over 2.5 / under 2.5 / over 3.5 goals, clean sheet, match result, draw and double chance. Each lists the matches of the next 48 hours with both clubs' record on that market, ranks the leagues, shows the clubs where it happens most and least, and gives the model's record as a table. Sections: `#today`, `#leagues`, `#clubs-ranked`, `#model-record`, `#what-it-means`. The nine addresses are in the index. - NFL hub: https://predictapp.io/en/nfl-predictions. Links the 32 teams, the league tables and the market guides. - NFL team: `/en/-predictions`, `/es/predicciones-`. Next game, results, spread, total, moneyline, team total, passing, rushing, receiving, touchdowns, defense by position. Sections: `#next-game`, `#results`, `#spread`, `#total`, `#moneyline`, `#team-total`, `#passing`, `#rushing`, `#receiving`, `#touchdowns`, `#defense`. - NFL player: `/en/-stats`, `/es/estadisticas-`, for quarterbacks, running backs, receivers and tight ends on a roster who played this season or last, or were drafted this year. This season, prop lines counted game by game, game log, role, last season, profile. - NFL fantasy: https://predictapp.io/en/fantasy-football-rankings, with a page per scoring format and per position and https://predictapp.io/en/fantasy-football-start-sit. Points scored so far, not a ranking of the week ahead. - NFL league tables: player props by stat (for example https://predictapp.io/en/nfl-receiving-yards-props) and defense against each position (for example https://predictapp.io/en/nfl-defense-vs-wr), all listed at https://predictapp.io/en/nfl-player-props. - The model's public record on finished matches: https://predictapp.io/en/record - Pricing: https://predictapp.io/en/pricing - API and MCP (English only): https://predictapp.io/en/api, https://predictapp.io/en/api/docs, https://predictapp.io/en/api/mcp ## What is not here - predictapp's prediction, probabilities and fair price for any match. Those are in the app, where a free account includes them for a match every day, and through the API and MCP server described above. - Betting odds, tips or picks. - Live scores. ## Names Club, competition and player names are used for identification only. predictapp is not affiliated with, endorsed by, or sponsored by any club, league, player or players' association.